User has Claude hand-edit Gemma neurons to tune Italian translation

Following MIT CSAIL's weight-space programming approach, a user had Claude Opus 5.5 rewrite 8 MLP neurons in Gemma 4 12B without training—about 92k of 11.9B parameters—to choose the Italian imperfect tense only in appropriate contexts.

• Study probes Qwen's 'Wait' reasoning loops2026-11-04
• Astra Natively Reads and Writes Weights to Improve Qwen2026-11-02
• Simon Willison: October 2026 is the vibe game dev inflection point2026-10-28
• Claude Opus 5.5 vs GPT-6 Astra: new AI models explained2026-10-20
• AI Cybersecurity Newsletter Roundup: Rogue Agents and Stolen Accounts2026-10-04